The one-sentence summary of my research interests is: I wish to evaluate and optimize human interaction technology and work processes to support team-based decision-making in safety-critical applications.  There are many concepts represented in this sentence, which are easiest to discuss in reverse order.  

Leveson (1986) defined safety-critical systems as: “complex, time-critical physical processes or mechanical devices, where a run-time error or failure could result in death, injury, loss of property, or environmental harm.”   Two examples of safety-critical situations that appear in my research are robot search and rescue (because robots need to quickly locate victims but avoid further injuring them) and air defense (because operators must quickly determine whether an unidentified aircraft is a “friend or foe”).  I am excited by both the importance and challenges of safety-critical domains that impose requirements for error-free and efficient interaction.  Besides the domains cited above I have worked with command and control systems (e.g., Drury and Cuomo, 1997), unmanned aerial vehicles (e.g., see Drury, Richer, Rackliffe, and Goodrich, 2006), and hazardous materials handling (Drury et al., 2008).

The air defense example illustrates the importance of providing technology, for example in the form of fused information displays, which supports efficient and accurate decision-making.  I have used cognitive task analysis methods such as Militello and Hutton’s (1998) Applied Cognitive Task Analysis technique to determine the major cognitive challenges for command and control center operators (Drury and Darling, 2008).  There is little consensus, however, on how to incorporate what we know about cognition into interaction designs.

There is even less consensus on how to best support team-based decision-making, even though many safety-critical systems are large-scale, multi-user systems that require some level of coordination or collaboration prior to making decisions and/or completing tasks.  One approach I have taken is to decompose the awareness needs of team members within a particular domain: both about each other and about the environment in which they are making decisions (the latter being usually called “situation awareness”).   For example, I decomposed team awareness needs, starting with the two overall categories of presence/identity and activities, in Drury and Williams (2002).  Simply having information isn’t sufficient for making good decisions, but in many real-life cases I have studied, people either do not have the necessary information, the information is not efficiently accessible, or is presented in a way that causes the user to understand it in a way other than what was intended.  Getting this part of the equation right is a pre-requisite for more sophisticated means of providing support for team-based decision-making.

Sometimes the right approach to aiding team-based decision-making is not to provide any new technology at all.  Rather, a team’s work processes may be suboptimal, resulting in decreased efficiency or accuracy.  I always try to observe people in their normal work environment, for example using the ethnographic technique known as Contextual Inquiry (Holtzblatt and Jones, 1993), to determine the combination of interaction technology and work processes that may be appropriate.  We used this technique with a military planning group (Swanson, Drury, and Lewis, 2004) and our recommendations included changing the format of an important daily meeting to make better use of the group’s time as well as the use of a tool already present in the group’s collaborative tool suite.

Regardless of whether new processes or new technologies are being proposed, evaluation methods are key to both ensuring that the user’s experience will be improved and that research progress has occurred.  It is usually more difficult to employ human-computer interaction techniques for multi-user, rather than single-user, systems (and especially synchronous, non-collocated systems) because of the challenge of running tests with multiple people in multiple places simultaneously (or simulating such conditions).  In my thesis work I developed the Synchronous Collaboration Awareness and Privacy Evaluation (SCAPE) method (Drury, 2001) to provide a fine-grained understanding of how well an application supports awareness of team members’ presence/identity and activities when such an understanding is appropriate, and team members’ privacy when the information needs to be protected.  Robots are also collaborative systems in the sense that even one robot and one human operator form a team; and I have also co-developed new evaluation methods for determining whether the robot interface provides operators with sufficient awareness (Drury et al., 2007).  In general, I enjoy the challenge of adapting or developing new methods for evaluating emerging technologies.

I characterize the best of my work as being foundational, interdisciplinary, and collaborative (in the sense of having been developed in conjunction with very talented colleagues).  For example, our decomposition of awareness needs for unmanned aerial vehicles (Drury, Riek, and Rackliffe, 2006) has provided a foundation upon which to build displays for these airborne robots.  Our taxonomy of human-robot interaction (Yanco and Drury, 2004) is widely cited and is often used in human-robot interaction courses.  I have used techniques, concepts, and approaches from HCI and computer-supported cooperative work when researching human-robot interaction. For example, we adapted HCI techniques to be more appropriate for use with robotics, resulting in different kinds of metrics for usability testing (Yanco, Drury, and Scholtz, 2004) and new “operators” (users’ actions or operations upon the system) for the Goals, Operators, Methods, and Selection rules (GOMS) method (Drury, Scholtz, and Kieras, 2007).  

Finally, I have benefited from researching and publishing with more than three dozen distinguished colleagues, including an MIT-trained roboticist, Prof. Holly Yanco of the University of Massachusetts Lowell; a co-developer of GOMS, Prof. David Kieras of the University of Michigan; and a renowned HCI metrics developer, Dr. Jean Scholtz of Pacific Northwest National Laboratory (formerly the National Institute of Standards and Technology).  A summary of my recent collaborations can be seen on my collaborations page.